Although especially applicable to “The Input/Loss Method” as installed at fossil-fired power plants, this invention may also be applied to any one of the “Input/Loss methods” installed at any thermal system burning a fossil fuel. Definitions for quoted terms are provided in the section entitled MEANING OF TERMS. The following paragraphs discuss prior art associated with The Input/Loss Method and with generic Input/Loss methods.
The principle background teachings of The Input/Loss Method are described in three patents: U.S. Pat. No. 6,584,429 which issued Jun. 24, 2003 and teaches a high accuracy method of determining boiler efficiency, hereinafter referred to as '429; U.S. Pat. No. 6,714,877 which issued Mar. 30, 2004 and teaches how effluent concentrations resultant from combustion may be corrected for errors, hereinafter referred to as '877; and, most importantly, U.S. Pat. No. 6,522,994 which issued Feb. 18, 2003 and teaches general methods of The Input/Loss Method. U.S. Pat. No. 6,522,994 originated as a PCT application resulting in the following patents: Canadian Patent 2325929; Australian Patent 762836; and European Patent (DE, GB, GR & IT) 1171834. These patents, U.S. Pat. No. 6,522,994, Canadian 2325929, Australian 762836 and European 1171834, are hereinafter collectively referred to as '994.
'994 is incorporated herein by reference in its entirety. '429 is incorporated herein by reference in its entirety. '877 is incorporated herein by reference in its entirety. In addition to '994, '429 and '877, a considerable technological foundation for The Input/Loss Method may be found in the following U.S. Pat. Nos.: 6,560,563, 6,651,035, 6,691,054, 6,745,152, 6,799,146, 6,810,358, 6,868,368 and 6,873,933. In addition, U.S. Pat. No. 7,039,555 was issued on May 2, 2006 with Terminal Disclaimers against U.S. Pat. Nos. 6,522,994 and 6,651,035. U.S. Pat. No. 7,039,555 principally teaches how tube failures in Recovery Boilers may be detected using The Input/Loss Method modified for sodium/hydrocarbon stoichiometrics.
Further still, a related pending application which again adds to the technology of The Input/Loss Method includes Canadian Patent Application No. 2479238 which stems from PCT/US02/37612 (WO2003/091881). The originating PCT application represented by PCT/US02/37612 resulted in U.S. Pat. No. 6,651,035 and European Patent (DE, GB, IE) 1502188, which both teach how tube failures in large steam generators may be detected using The Input/Loss Method. U.S. Pat. No. 6,651,035 was originally filed as a U.S. Continuation-In-Part to an application which became U.S. Pat. No. 6,745,152.
One of the Input/Loss methods, a rudimentary method, is described in U.S. Pat. No. 5,367,470 which issued Nov. 22, 1994 (with Dec. 14, 1989 priority), and in U.S. Pat. No. 5,790,420 which issued Aug. 4, 1998. U.S. Pat. No. 5,790,420 was originally filed as a U.S. Continuation-In-Part to an application which became U.S. Pat. No. 5,367,470.
Other known Input/Loss methods are thoroughly discussed in the BACKGROUND OF THE INVENTION section of '994; this discussion is referenced herein as being important.
For many years the energy industry has attempted to categorize coals. Although there are four major ranks of coal in the U.S. classification scheme (anthracite, bituminous, sub-bituminous and lignite), these have been sub-divided by ASTM D388, “Standard Classification of Coals by Rank”. Refer to TABLE B1 for ASTM D388 categories (an incorrect energy conversion was used in this standard, 2.3255 kJ/kg/Btu/lb, versus 2.3260 kJ/kg/Btu/lb). One problem immediately seen in TABLE B1 is its lack of specificity, ASTM D388 basically employs either As-Received calorific values, and/or proximate analyses on a dry basis to judge coals. “Ultimate Analysis” data is not employed. Higher Rank coals are classified according to fixed carbon on a dry basis while the lower Rank coals are classified by As-Received calorific value (wet basis). Figure X1.1 of ASTM D388 presents a typical single-variant correction between weight fraction of volatile matter and Reflectance in oil. A general discussion of coal classifications may be found in the text The Chemistry and Technology of Coal by J. G. Speight, Marcel Dekker, Inc, New York & Basel, which discusses coal classifications in Chapter 1 (pages 3-19), elemental analysis on pages 83-84 and evaluation techniques in Chapter 8 (pages 165-199). Note that examples of single-variant analyses are presented in this text's FIGS. 1.2, 8.10, 8.11 and 8.12; several of these displaying weight fraction of fuel hydrogen versus weight fraction of fuel carbon. As seen, these plots represent only broad-brush correlations, hardly capable of supporting any of the Input/Loss methods.
TABLE B1ASTM Classification by RankRank (abbreviation)Characteristicsmeta-anthracite (ma)Fixed carbon ≧98%.anthracite (an)Fixed carbon ≧92% and <98%.semi-anthracite (sa)Fixed carbon ≧86% and <92%.low volatile bituminous (lvb)Fixed carbon ≧78% and <86%.medium volatile bituminous (mvb)Fixed carbon ≧69% and <78%.high volatile A bituminous (hvAb)CV ≧ 14000 Btu/lb(CV ≧ 32557 kJ/kg), withFixed carbon <69%high volatile B bituminous (hvBb)14000 Btu/lb > CV ≧ 13000 Btu/lb(32557 kJ/kg > CV ≧ 30232 kJ/kg)high volatile C bituminous (hvCb)13000 Btu/lb > CV ≧ 10500 Btu/lb(30232 kJ/kg > CV ≧ 24418 kJ/kg)sub-bituminous A (sub A)11500 Btu/lb > CV ≧ 10500 Btu/lb(26743 kJ/kg > CV ≧ 24418 kJ/kg)sub-bituminous B (sub B)10500 Btu/lb > CV ≧ 9500 Btu/lb(24418 kJ/kg > CV ≧ 22090 kJ/kg)sub-bituminous C (sub C)9500 Btu/lb > CV ≧ 8300 Btu/lb(22090 kJ/kg > CV ≧ 19300 kJ/kg)lignite A (lig A)8300 Btu/lb > CV ≧ 6300 Btu/lb(19300 kJ/kg > CV ≧ 14650 kJ/kg)lignite B (lig B)6300 Btu/lb > CV(14650 kJ/kg > CV)
There are seemingly as many coal categories used in Europe as countries. In general, Europeans categorize coal as either hard or soft depending on ash-free calorific value. Sub-groups are then classed by volatile matter, coking properties, etc. resulting in a complex three-digit numbering system. No European system employs Ultimate Analysis data to classify coals, at best proximate analyses are employed. Refer to “Brown Coals and Lignites—Classification by Types on the Basis of Total Moisture Content and Tar Yield”, International Organization for Standards, ISO 2950-1974(E). The broad categorizations favored in Europe (i.e., hard or soft coals) is also reflected in the so-called van Krevelen diagrams whose use dates from 1950. Van Krevelen diagrams are plots of atomic hydrogen/carbon versus atomic oxygen/carbon ratios. This and related research is summarized in the books: DW van Krevelen and J Schuyer, “Coal Science, Aspects of Coal Constitution”, Elsevier Science Publishers, Amsterdam, 1957; and DW van Krevelen, “Coal: Typology, Physics, Chemistry, Constitution”, Third Edition, Elsevier Science Publishers, Amsterdam, 1993. If oxygen may be held constant, then van Krevelen digrams reduce fundamentally to hydrogen versus carbon relationships found useful when applying '994 techniques.
It is also useful to recognize that the analysis of fossil fuels may be accomplished using the Excel® computer program. Excel is owned by the Microsoft Corporation, Redmond, Wash. state in the U.S. Excel is a registered trademark of Microsoft Corporation. Fossil fuel data is typically obtained as Ultimate Analysis data with As-Received fuel water, fuel ash and calorific values. As used to develop this invention, and used throughout its presentation herein, such data was analyzed using Excel. All “R2 values” mentioned herein, commonly termed the Coefficient of Determination, have been computed by Excel using regression analysis. Excel's R2 value represents the percent variation in a y-variable that is explained by the independent x-variable. Only linear regression was used herein. There are classical problems associated with R2 values as are well known to one skilled in statistics. One such problem, and one important to this invention, is evident when data presents an even scatter about a linear mean. Such a situation might lead to a high R2 value which does not truly reflect a y-variable being predictable by the independent x-variable (simply put, the R2 value may appear acceptable, but the functionality is too coarse to be useable). The most straightforward method to address such situations is to simulate data patterns associated with their end use and to then evaluate the direct impact their variances have on computed output. For example, the impact on The Input/Loss Method's computed calorific value of a 1.0% variance in predicted fuel carbon (and thus affecting computed calorific value) may be assessed most conservatively by assuming a 1.0% variance in effluent CO2; such a 1.0% variance may be observed, and verified, from plotted data. Another method of evaluating distributed data patterns is to simply apply engineering judgement by looking at the plots: they are either unreasonable or portend fundamental understanding with obvious certainty.
The technologies underwriting The Input/Loss Method, witnessed by the aforementioned patents and patent applications, were based on recognizing that if the effluent concentrations from combustion are used to determine fuel chemistry, then fundamentally more unknowns are involved than practical equations are available. '994 presented a solution to this problem by teaching that fuel hydrogen may have a functional relationship with fuel carbon; see Eq.(45) in '994 and the definition of “reference fuel characteristics” in '994. Other relationships are fuel oxygen versus fuel carbon, and fuel nitrogen versus fuel carbon; refer to Eqs.(43) and (44) in '994 and associated discussion above Eq.(42) in '994. For example, the correlation constants A5 & B5 used in Eq.(45) in '994 derive directly from ultimate analysis data, for example, as seen in FIG. 3 of '994. Eq.(42) in '994 presents an explicit solution to moisture-ash-free (MAF) molar fuel carbon employing correlation coefficients: A3 & B3 from MAF molar fuel oxygen as a function of MAF molar fuel carbon of Eq.(43) in '994; and A5 & B5 from MAF molar fuel hydrogen as a function of MAF molar fuel carbon of Eq.(44) in '994. These correlations provided the missing equations. They all are simple single-variant molar correlations using hydrogen versus carbon, or oxygen versus carbon; e.g., the single-variant is molar hydrogen as observed in '994 Eqs.(45). There was no other known art or techniques for solving the underlying problem.
Wherein The Input/Loss Method has been installed at a number of power plants, certain situations have arisen in which single-variant relationships such as fuel hydrogen versus fuel carbon are simply not adequate. This has been found true when employing “reference fuel characteristics” as defined and taught in '994. It has been found that this situation is especially true if dealing with the following fuel types: Irish peat; Powder River Basin coals; and what is termed “High Seas” coal. Irish peat is of importance as it represents a typical indigenous fuel source, not only for the Republic of Ireland, but also for Poland, for Finland and for Minnesota in the U.S. Peat's dry chemistry may vary considerably given its haphazard formation as immature coal, and its fuel water content typically varies wildly. The MAF characteristics of peat are not unlike lignite found in Texas, Australia and Greece. Powder River Basin coals have an enormous, and growing, financial impact on the United States and Canada as it represents the largest single source of coal fuel being fired in North American power plants. Over 120 power plants use Powder River Basin coals, growing by some estimates at 15%/year. Powder River Basin coals have low sulfur concentrations, but are high in fuel water with highly variable fuel chemistries reflecting over a dozen mines located in several western states in the U.S. High Seas coal is defined as high energy coal which is frequently bought, literally, while coal-carrying cargo ships are on the high seas. It may be categorized, as high volatile bituminous coal. High Seas coal typically has low fuel water, but fuel chemistries reflecting variability associated with world-wide sourcing. High Seas coal typically has calorific values in the range of 25,586 to 31,401 kJ/kg (11,000 to 14,000 Btu/lbm). There are other fuels which, it is anticipated, will receive higher interest over the coming years, but which will have similar variabilities. One such fuel is switch grass, grown in the U.S. as an environmentally friendly (and renewable) fossil fuel. Another, is wood waste (i.e., bio-mass fuel), being burned in the western states of the U.S.
If Irish peat, Powder River Basin coals and High Seas coals were not significantly used, then the method taught in '994 would be adequate given a supposed well-characterized fuel. By well-characterized is meant that needed correlations (e.g., MAF molar fuel hydrogen as a function of MAF molar fuel carbon) have R2 values which exceed 90%. Note however that if an R2 value at 90% is considered inadequate (versus, say 98%), or not, the practical application of '994 was, indeed, limited to this level of predictability as a direct consequence of simple single-variant correlations.
It is important to note that “reference fuel characteristics”, as defined in '994, represents a taught procedure, one in which hydrogen versus carbon relationships are developed based on historical fuel data. It does not specify usable data. When the method of '994 was installed in PRB burning powers plants, coal from specific regions within the Basin would require characterization. The Boardman Coal Plant, operated by Portland General Electric and using The Input/Loss Method, was characterized specifically to PRB Decker coal. The Nebraska City Unit 1, operated by Omaha Public Power District and using The Input/Loss Method, was characterized specifically to PRB Caballo Rojo coal. And the same even for Irish peat. The Lough Ree Power Station, operated by the Electricity Supply Board and using The Input/Loss Method, was characterized specifically to Irish peat found near Lanesboro, Ireland, although the West Offlay Power Station, also burning Irish peat, not 56 km (35 miles) away, was characterized specifically to the Shannonbridge region. '994 taught a procedure requiring historical data, requiring unique reference fuel characteristics to be programmed in a computer for each installation. What is needed is a generic method such that a single procedure satisfies an entire Rank of coal, without routine need of historical data. At the time of '994 there was no other known art. When considering variable fuels, as defined by poor R2 values resultant from using simple single-variant correlations, the '994 method has not proven to be generic as it suffers from a lack of flexibility under certain circumstances.
The databases of Ultimate Analyses and calorific values used to develop this invention derive from the following sources: 1) Pennsylvania State University, Organic Petrology Laboratory database containing over 1200 Ultimate Analyses and associated calorific values from over 400 mines; 2) Powder River Basin coal data containing approximately 250 samples from 19 different regions within the Basin; 3) so-called High Seas coal data containing 320 samples from over 50 mines from 14 states in the U.S., South Africa, Poland, Russia and Colombia, this data includes numerous spot analyses obtained from power plants actually using such coal (i.e., from the Moneypoint station, Republic of Ireland, from the Brandon Shores station, Maryland state in the U.S., and from the Jorf Lasfar station, Morocco); and 4) Irish peat data containing approximately 160 samples from 6 different regions within the Republic of Ireland, notably the data having been collected over a considerable time period, from 1963 through 2005. In total the analyzed data consisted of approximately 1930 Ultimate Analyses and corresponding calorific values.
As seen in FIG. 1 for Irish peat, as seen in FIG. 3 for Powder River Basin coals, and as seen in FIG. 5 for High Seas coal the ability of '994 technology to reasonably provide functionality between MAF molar fuel diatomic hydrogen versus MAF molar fuel carbon is wanting, as based on simple single-variant correlations. For the Irish peat data of FIG. 1, the R2 value was found at 65.90%. For the Powder River Basin coal data of FIG. 3, the R2 value was found at 71.93%. For the High Seas coal data of FIG. 5, the R2 value was found at 81.77%. Note, that although these fuels are not well-characterized using single-variant correlations, their industrial use is quite real; such use demands an improved approach. It also must be noted that a poor R2 value for MAF molar fuel hydrogen versus MAF molar fuel carbon, portents an even poorer R2 value for fuel oxygen versus fuel carbon; and poorer yet for fuel nitrogen versus fuel carbon. For MAF molar fuel oxygen versus MAF molar fuel carbon, the R2 values were found at 36.48% for Irish peat, 14.01% for Powder River Basin coals and 64.23% for High Seas coals. Such non-predictability results forced the user of '994 technology, for these types of fuels, to assume that MAF molar fuel oxygen be keep constant. As an example of the practical problem, the typical power plants using High Seas coal (e.g., Moneypoint, Brandon Shores and Jorf Lasfar) do not sort the fuel, they burn whatever is on the loading docks having acquired the fuel from anywhere in the world based on price, etc. An improvement of methods is needed if such fuels are to be described with sufficient predictability for Input/Loss methods to function with the high accuracy of which it is capable. In summary the following features associated with '994 methods have proven to be inadequate:                its use of “reference fuel characteristics”, as defined in '994, employing single-variant correlations and its use of the L5 Factor;        “reference fuel characteristics”, as defined in '994, require historical data;        poor R2 values (<90%) for the important MAF molar fuel hydrogen versus MAF molar fuel carbon relationships and very poor R2 values (<70%) for oxygen versus carbon relationships which results in forcing MAF molar fuel oxygen to be held constant;        the use of equations which solve for elemental constituents which combine single-variant correlation constants and stoichiometric terms;        assuming fuel nitrogen is constant; and        the use of numerical minimum and maximum limits applied to fuel concentrations as taught being a portion of the “reference fuel characteristics” defined in '994, has caused inconsistencies (as seen in FIG. 1, FIG. 3 and FIG. 5, a maximum αMAF-4 implies a minimum αMAF-5, and typically a minimum αMAF-3, thus the MAF summation could lead to inconsistencies which is an intrinsic disadvantage of single-variant analysis).        
As demonstrated in FIG. 1, FIG. 3 and FIG. 5, the method taught in '994 simply cannot produce R2 values near 98% for many important fuels without specialized study. If the fossil fuel is well characterized, and especially if the coal is of a higher Rank and having low fuel oxygen (e.g., anthracite, semi-anthracite and sub-bituminous A) the method of '994 using single-variant correlations may produce R2 values near 90%. However, if to reach predictability values at the 98% level, understanding the genesis of fossil fuels is required. It requires a clear inventive step beyond the established technology of '994. There is no known art which addresses fundamental fossil fuel genetics such that R2 values at the 98% level might be achieved, at least for the majority of commercial fuels. Other than '994 and other Input/Loss methods discussed in the BACKGROUND OF THE INVENTION section of '994, there is no established art directly related to this invention. There is a clear need for an apparatus supported by a methodology which describes fossil fuel genetics in such a manner that fuel chemistry may be resolved and/or confirmed.